Separate Cyclic AMP Sensors for Neuritogenesis, Growth Arrest, and Survival of Neuroendocrine Cells*

Background: Three cAMP sensors (PKA, Epac1/2, and NCS/Rapgef2) coexist in neuroendocrine cells. Their roles in differentiation require elucidation. Results: Epac2, PKA, and NCS/Rapgef2 independently gate signaling for growth arrest, cell survival, and neuritogenesis after GPCR-Gs engagement in PC12 cells. Conclusion: Parallel, insulated pathways effect cAMP-dependent neuroendocrine cell differentiation. Significance: Assays for parcellated cAMP signaling in neuroendocrine cells have a broad application for CNS drug discovery. Dividing neuroendocrine cells differentiate into a neuronal-like phenotype in response to ligands activating G protein-coupled receptors, leading to the elevation of the second messenger cAMP. Growth factors that act at receptor tyrosine kinases, such as nerve growth factor, also cause differentiation. We report here that two aspects of cAMP-induced differentiation, neurite extension and growth arrest, are dissociable at the level of the sensors conveying the cAMP signal in PC12 and NS-1 cells. Following cAMP elevation, neuritogenic cyclic AMP sensor/Rapgef2 is activated for signaling to ERK to mediate neuritogenesis, whereas Epac2 is activated for signaling to the MAP kinase p38 to mediate growth arrest. Neither action of cAMP requires transactivation of TrkA, the receptor for NGF. In fact, the differentiating effects of NGF do not require activation of any of the cAMP sensors protein kinase A, Epac, or neuritogenic cyclic AMP sensor/Rapgef2 but, rather, depend on ERK and p38 activation via completely independent signaling pathways. Hence, cAMP- and NGF-dependent signaling for differentiation are also completely insulated from each other. Cyclic AMP and NGF also protect NS-1 cells from serum withdrawal-induced cell death, again by two wholly separate signaling mechanisms, PKA-dependent for cAMP and PKA-independent for NGF.

An array of extracellular signaling molecules, including hormones and neurotransmitters, cause cellular changes by regulating levels of the intracellular second messenger cAMP. In the central nervous system, regulation of intracellular cAMP has been shown to control intracellular processes underlying synaptic plasticity and memory formation (1), guide axonal elongation (2), and support neuronal survival in both the developing and adult brain (3). PC12 cells isolated from rat adrenal pheochromocytoma have long been used as a model of neural devel-opment and differentiation. Both receptor tyrosine kinaseactivating neurotrophins, such as NGF, and neuropeptides that elevate intracellular cAMP, such as pituitary adenylate cyclase-activating polypeptide (PACAP) 2 , cause PC12 cells to undergo the dissociable processes of morphological differentiation (neurite outgrowth) and growth arrest (4,5) that, together, comprise differentiation.
Elevation of cAMP by GPCR signaling leads to activation of multiple downstream cAMP-responsive proteins (cAMP sensors). Protein kinase A (PKA) is the best characterized cAMP sensor. However, additional proteins functioning as cAMP sensors have since been identified. These include the exchange proteins activated by cAMP (Epacs) (6,7) and the neuritogenic cAMP sensor (NCS) Rapgef2 (8 -10). Each of these three cAMP sensors (PKA, Epac, and NCS/Rapgef2) have been implicated in some aspect of PC12 cell differentiation (8,11,12). However, it has not been determined whether each cAMP sensor fulfills a particular set of cellular tasks in response to cAMP elevation. Furthermore, it remains unknown whether the three cAMPinduced signaling pathways cross-regulate each other or whether they act as functionally insulated signaling channels.
The discovery of NCS/Rapgef2 as the cAMP sensor mediating neuritogenesis in response to PACAP and the use of NS-1 cells as a PC12 cell subclone in which differentiation events can be conveniently measured and precisely quantified has provided new pharmacological and cell biological opportunities to define cAMP signaling parcellation. Here, we have determined the downstream signaling pathways engaged by cyclic AMP to cause both differentiation (neurite extension and growth arrest) and cell survival upon serum withdrawal. We report that three separate cAMP sensors independently mediate these three effects of cAMP elevation and, further, that the actions of NGF, while using the same final cellular effectors for neuritogenesis and growth arrest as cAMP elevation, do so without the need for signaling through cAMP.
Cell Culture-Neuroscreen-1 (NS-1) cells are a subclone of PC12 cells purchased from Cellomics. All solutions used for cell culture were purchased from Invitrogen unless noted otherwise. NS-1 cells were cultured in RPMI 1640 medium supplemented with 10% horse serum (HyClone), 5% heat-inactivated fetal bovine serum, 2 mM L-glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin. Cells were grown in flasks (Techno Plastic Products) coated with collagen type I from rat tail, as described previously (9), at 37°C in a humidified incubator containing 5% CO 2 . PC12 cells (PC12-G, see Ref. 22) were cultured in DMEM supplemented with 7% horse serum, 7% heat-inactivated fetal bovine serum, 25 mM HEPES, 2 mM L-glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin. PC12 cells were grown at 37°C in a humidified incubator containing 5% CO 2 and, for experiments, were plated on poly-Dlysine-coated 6-well plates as described previously (23). All cells routinely tested negative for mycoplasma and were used between passages 5 and 16 for the experiments reported here.
Neurite Outgrowth Measurements-NS-1 cells were dispensed into 6-well plates, and, the following day, media were changed to media containing drugs. Cells were treated for 48 or 72 h, as indicated. Using a ϫ20 lens, photomicrographs were then acquired with a computer-assisted microscope. Images were randomized, and a blinded observer counted the number of cells, the number of neurites, and the length of the neurites in each field using NIS-Elements BR (Nikon).
Proliferation Assays-3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) was purchased from Sigma and was kept in the dark at all times. Stocks (1 mg/ml) were prepared in culture media and stored for 2 weeks or less at 4°C. NS-1 cells were plated in 96-well plates (1 ϫ 10 4 cells/well) and treated accordingly for varying lengths of time. Media were then aspirated and replaced with MTT (0.2 mg/ml) in culture media. Cells were exposed to MTT for 90 min in the dark at 37°C. MTT-containing media were then aspirated, and formazen salts were extracted from each well with 70 l of DMSO. The optical density of the extracted formazen product, proportional to the number of viable cells (24), was then measured using a microtiter plate reader (PerkinElmer Life Sciences).
Western Blotting-Cells were seeded in 12-well plates and grown overnight. Cells were pretreated with small molecule inhibitors for 30 min prior to the addition of agonists. For measurements of ERK and p38, cells were treated with agonists for 10 min, and then media were aspirated and cells were collected in ice-cold lysis buffer (150 mM NaCl, 50 mM Tris-HCl, 1% Nonidet P-40, and 1 mM EDTA) containing Halt protease and phosphatase inhibitors (Pierce Biotechnology). Protein concentrations were normalized, and samples for electrophoresis were diluted in LDS sample buffer (Invitrogen) to a final protein concentration of 1 g/l. Proteins (20 g/lane) were then sep-arated by SDS-PAGE on 4 -12% polyacrylamide Bis Tris gels (Invitrogen). Gels were blotted onto nitrocellulose membranes (Invitrogen) using a semidry transfer apparatus (Invitrogen) at 30 V for 2 h at room temperature. Membranes were then blocked with 5% skim milk dissolved in Tris-buffered saline with 1% Tween 20 (TBST) for 2 h. Membranes were incubated overnight at 4°C with the following primary phospho-specific antibodies, which were purchased from Cell Signaling Technology: phospho-ERK (recognizes dual-phosphorylated ERK1 and ERK2, catalog no. 4370) or phospho-p38 (recognizes dualphosphorylated p38␣, ␤, ␦, and ␥, catalog no. 4511). Membranes were then washed five times in TBST and incubated with appropriate HRP-coupled secondary antibodies (Pierce) in blocking buffer for 1 h. Membranes were again washed five times in TBST and exposed to a chemiluminescent HRP substrate (Super Signal West Pico, Pierce Biotechnology). Membranes were imaged using a cooled charge-coupled device camera (Alpha Innotech). Each membrane was then stripped for 15 min using Restore Plus Western blot stripping buffer (Pierce), washed five times in TBST, incubated in blocking buffer for 1 h, and reprobed with primary antibodies raised against total ERK (Cell Signaling Technology, catalog no. 4695) or p38 (Cell Signaling Technology, catalog no. 8690).
G 1 /G 0 Biosensor Measurements-An LNCX retroviral vector expressing HDHBc-tdimer2 was generated as described previously (23). PC12 cells transduced with the LNCX-HDHBc-tdimer2 retroviral vector were cultured in selection medium containing G418 (800 g/ml). Cells were plated in poly-D-lysine-coated 6-well plates, grown overnight, and treated as indicated. At the end of the experiment, cells were photographed with a Nikon microscope using a ϫ20 objective with a rhodamine filter. All cells were counted, and those with nuclearly confined red fluorescence were counted as cells in G 1 /G 0 phase, whereas those with diffuse cytoplasmic red fluorescence were counted as being in S/G 2 phase.
Silencing of Epac1 and Epac2-shRNA for rat Epac1 (target sequence aggaaccgatatacggtaa) and Epac2 (target sequence caagaaacacgaagcgtat) were expressed in psi-Lv-HIVH1 lentiviral vectors (GeneCopoeia). Lentiviral particles were generated by cotransfecting HEK293T cells, grown in 10-cm dishes using the ProFection calcium phosphate system (Promega) according to the instructions of the manufacturer with plasmids encoding shRNA expression vectors (10 g), gag-pol-rev (6.5 g), and vesicular stomatitis virus (3.5 g). 48 h post-transfection, supernatants were harvested, filtered through 0.45-m filters, and applied to cultures of 60 -70% confluent NS-1 cells growing in T25 flasks. 12 h post-transduction, NS-1 cells were split and cultured in selection medium containing 1 g/ml puromycin. Following two passages in selection medium, Ͼ98% of transduced NS-1 expressed visible GFP upon epifluorescence illumination. Knockdown of Epac1 and Epac2 protein abundance was confirmed by Western blotting as described above using antibodies from Cell Signaling Technology (catalog nos. 4155 and 4156).
Measurements of Rap1 Activation-Rap1-GTP was measured using the Active Rap1 pulldown and detection kit (Pierce, catalog no. 16120) according to the instructions of the manufacturer. NS-1 cells were grown to near confluency in 6-well plates. Cells were treated for 10 min as indicated and were then lysed in ice-cold lysis buffer provided by the manufacturer. An aliquot of protein from each sample (500 g) was incubated in a solution of 20 g of GST-RalGDS-RBD in a glutathione resin slurry for 1 h. Samples were then centrifuged through spin cups, washed three times with lysis buffer, dissolved in reducing sample buffer, vortexed, and heated to 95°C for 5 min. Samples were then analyzed by Western blotting as described above and probed with an antibody raised against Rap1 at a dilution of 1:1000 (Pierce). To account for possible differences in Rap1 content between samples, unpurified protein samples (20 g) that corresponded to affinity-purified samples were also analyzed by Western blotting on separate gels using the same procedure.
Propidium Iodide Staining-Propidium iodide (PI) is a dye that fluoresces upon binding to DNA and is highly polar. Therefore, it enters and stains only cells with damaged membranes. Stocks of PI (5 mg/ml) were prepared in PBS and stored at 4°C protected from light for up to 1 month. Following treatment, media were aspirated, and cells were exposed to 5 M PI in PBS for 30 min in the dark at 37°C. Fluorescence was then viewed using an inverted fluorescence microscope using a rhodamine filter. PI-positive and PI-negative cells were then counted manually by a blinded observer using NIS Elements (Nikon).
Calculations and Statistics-All statistical analyses were performed using Sigma Plot (Systat). For datasets with a Gaussian distribution, statistical comparisons of multiple groups within an experiment were made using analysis of variance followed by Bonferroni-corrected t tests comparing each condition to controls. In experiments where data were not normally distributed, data were analyzed by Kruskal-Wallis non-parametric analysis of variance followed by Dunnet's or Dunn's post hoc tests comparing treated groups to controls. For dose-response experiments, curves were fit to dose-response data using four-parameter logistic regression where appropriate.

RESULTS
We reported previously that intracellular cAMP, acting at NCS/Rapgef2, causes neurite extension (neuritogenesis) in NS-1 cells. NCS/Rapgef2 enhances GTP loading on the small G protein Rap1, allowing its association with B-Raf, thus activating MEK and ERK (8). This pathway is activated by the neuropeptide PACAP through interaction with the GPCR PAC 1 and subsequent G s -dependent stimulation of adenylate cyclase and elevation of cAMP (8,9). NGF also stimulates both neurite elongation and growth arrest. ERK is necessary for neuritogenesis because of either cAMP or NGF, and therefore we wished to see whether cAMP and NGF share a common pathway for inducing either neuritogenesis or growth arrest.
NGF and cAMP Stimulate Neuritogenesis via Separate Signaling Pathways-NS-1 cells were differentiated by treatment for 48 h with the lipophilic cAMP analog 8-CPT-cAMP (100 M) or NGF (100 ng/ml). As seen in Fig. 1, A-C, treatment with either 8-CPT-cAMP or NGF resulted in both neurite extension and growth arrest. NGF signaling for differentiation has been reported to involve formation of a TrkA multiprotein complex containing NCS/Rapgef2 (also called PDZ-GEF1) and requiring activation of Rap1 (25). To see whether cAMP interaction with NCS/Rapgef2 is necessary for NGF-induced differentiation, we treated NS-1 cells with the inhibitor of adenylate cyclase and NCS/Rapgef2, SQ22,536 (26), and then challenged cells with either 8-CPT-cAMP or NGF. As seen in Fig. 1, A-C, SQ22,536 blocked cAMP-dependent neurite extension while not affecting NGF-dependent signaling or cAMP-induced growth arrest.
It has also been reported that cAMP or one of its downstream effectors signals via transactivation of TrkA receptors (20). We wished to determine whether cAMP-induced ERK activation and neurite extension may involve transactivation of TrkA receptors. NS-1 cells were treated with either 8-CPT-cAMP (100 M) or NGF (100 ng/ml) in the absence or presence of 200 nM of the TrkA inhibitor K-252a (16). K-252a significantly blocked NGF-induced ERK activation while not affecting cAMP-induced activation of ERK (Fig. 1, D and E). In the same experiments SQ22,536 blocked cAMP-dependent but not NGF-dependent activation of ERK (Fig. 1, D and E). Following the same pharmacological profile as observed for ERK activation, NGF-induced neuritogenesis was blocked by K-252a, whereas cAMP-dependent neurite outgrowth was unaffected by the addition of K-252a (Fig. 1, F and G). Together, these data indicate that the signaling pathways for neuritogenesis activated by NGF and cAMP are functionally insulated insofar as cAMP-initiated neuritogenesis does not require TrkA receptor activation and that NGF-initiated neuritogenesis does not require NCS/Rapgef2.
Epac Mediates cAMP-dependent Growth Arrest-NS-1 cells also undergo growth arrest when treated with NGF or cAMP. NCS/Rapgef2, although necessary for neuritogenesis, is apparently not required for cAMP-dependent growth arrest (Fig. 1, A and C). To determine the mechanism through which cAMP mediates growth arrest in NS-1 cells, the proliferation rate of NS-1 cells grown in medium containing 8-CPT-cAMP (100 M) was monitored by estimation of cell number at the indicated intervals with the MTT assay. After 3 days of treatment, the effects of 8-CPT-cAMP (100 M) on growth arrest were evident ( Fig. 2A). To determine which cAMP sensor mediates growth arrest, NS-1 cells were treated with 8-CPT-cAMP (100 M) for 3 days with varying concentrations of H-89 (a PKA inhibitor), SQ22,536 (a NCS/Rapgef2 inhibitor), or ESI-09 (an Epac inhibitor). As seen in Fig. 2B, ESI-09 inhibited growth arrest because of 8-CPT-cAMP, whereas neither H-89 nor SQ22,536 significantly inhibited growth arrest, suggesting that Epac is necessary for cAMP-dependent growth arrest. Consistent with the notion that Epac underlies cAMP-dependent growth arrest, the Epac-selective cAMP analog 8-CPT-2Ј-O-Me-cAMP (27) caused growth arrest to a similar extent as 8-CPT-cAMP, which activates all three cAMP sensors (Fig. 2C).
Signaling through Epac Causes Growth Arrest in a Rap1-independent Manner-Cyclic AMP-induced neuritogenesis in NS-1 cells requires NCS/Rapgef2-mediated stimulation of Rap1 (8). Rap is also the best characterized effector of Epac signaling. Therefore, we wished to determine whether Epacinduced growth arrest is Rap-dependent. As seen in Fig. 3A Fig. 3, D and E, FTS-A inhibited cAMP-dependent neuritogenesis, indicating that it remains biochemically active during the extended treatment. Furthermore, although FTS-A blocked neurite extension in these experiments, it failed to inhibit cAMP-dependent growth arrest (Fig. 3, C and E). Together, these data suggest that cyclic AMP and Epac induce growth arrest via a Rap1-independent signaling pathway. Therefore,  we examined other possible downstream effectors of Epacmediated signaling.
The MAP Kinase p38 Is Necessary for Epac-dependent Growth Arrest-ERK is necessary for PC12 cell neuritogenesis (9,29,30), and ERK has also been shown to directly mediate growth arrest in certain cell types, such as transformed fibroblasts (31). To investigate a possible role for MEK/ERK in growth arrest, we treated NS-1 cells with 8-CPT-cAMP (100 M) in the absence or presence of varying concentrations of the MEK inhibitor U0126. As seen in Fig. 4A, U0126 failed to inhibit the effect of 8-CPT-cAMP on growth arrest, suggest-ing that ERK is not a component of the cAMP growth arrest pathway. Activation of the MAP kinase p38 has been shown to contribute to growth arrest through multiple mechanisms (32). Moreover, Epac has been shown to activate p38 in cerebellar and hippocampal neurons (33,34). Accordingly, we measured cAMP-induced p38 activation using antibodies that specifically recognize p38 phosphorylated at Thr 180 and Tyr 182 . As seen in Fig. 4B, treatment with either the panspecific cAMP analog 8-CPT-cAMP or the Epac-selective 8-CPT-2Ј-O-Me-cAMP increased the abundance of phosphorylated p38. Similar results were obtained using PC12  (Fig. 4C).
We next established that the p38 inhibitor SB 239063 (10 M) effectively blocked Epac-dependent p38 activation, whereas U0126 (10 M) had no effect on NS-1 cells (Fig. 4D). Similarly, in PC12 cells, pretreatment with SB 239063 effectively inhibited p38 phosphorylation occurring in response to treatment with 100 M 8-CPT-2Ј-O-Me-cAMP (Fig. 4E). To determine whether p38 is required for Epac-dependent growth arrest, NS-1 cells were treated with either 8-CPT-cAMP or 8-CPT-2Ј-O-Me-cAMP for 3 days in the absence or presence of varying concentrations of SB 239063. As seen in Fig. 4F, SB 239063 potently inhibited 8-CPT-2Ј-O-Me-cAMP-induced growth arrest with an IC 50 value of 460 ϩ 10 nM.
The Neuropeptide PACAP Causes Growth Arrest via Epac/ p38 Activation-Because cyclic nucleotide analogs cause NS-1 cells to undergo growth arrest via a signaling pathway including Epac and p38, we wished to see whether a GPCR ligand could also activate this signaling pathway. The neuropeptide PACAP plays protean roles in the development of the nervous system (35) and is well known to cause differentiation in PC12 cells (36). As seen in Fig. 5A, treatment with PACAP (100 nM) caused growth arrest to a comparable extent as cAMP analogs. PACAP-induced growth arrest was also potently inhibited by the p38 inhibitor SB 239063 (IC 50 ϭ 710 Ϯ 40 nM), suggesting that PACAP-induced growth arrest is also p38-dependent.
We next wished to see whether PACAP activates p38 and, if so, whether it also occurs in an Epac-dependent manner. NS-1 cells were treated with PACAP (100 nM) for 10 min with or without ESI-09 (10 M). As seen in Fig. 5, B and C, PACAP caused a significant increase in phosphorylated p38. The effect of PACAP was inhibited significantly by the addition of ESI-09, indicating that Epac is required for PACAP-dependent p38 activation. PACAP-dependent p38 phosphorylation was also blocked effectively by ESI-09 (10 M) in PC12 cells (Fig. 5D).
To confirm that the effects of PACAP on growth arrest are mediated by Epac, NS-1 cells were treated with either PACAP (100 nM) or 8-CPT-2Ј-O-Me-cAMP (100 M) for 3 days in the Note that all four isoforms of (phospho)p38, ␣, ␤, ␥, and ␦, are recognized by the pp38 antibody employed here. The signals detected here are from pp38␣, ␤, or ␦ or a combination thereof (apparent molecular weights of these isoforms, ϳ40 -43 kDa) but not pp38␥ (apparent molecular weight, ϳ46 kDa).
absence or presence of ESI-09 (10 M). As seen in Fig. 5E, both PACAP and 8-CPT-2Ј-O-Me-cAMP caused growth arrest. The effects of both were significantly inhibited by ESI-09, suggesting that PACAP, indeed, causes growth arrest of NS-1 cells via engagement of the Epac pathway.
To determine which Epac is required for cAMP-dependent growth arrest, Epac1 and Epac2 were each stably silenced in NS-1 cells using shRNA. Cell lines in which the abundance of the target protein product was reduced by Ն90% relative to that in untransduced controls were selected for further experiments (Fig. 5F). Parental NS-1 cells as well as lines deficient in Epac1 and Epac2 were treated for 3 days in the absence or presence of cyclic AMP-elevating/mimicking agents PACAP-38 (100 nM), forskolin (25 M), 8-CPT-cAMP (100 M), or the receptor tyrosine kinase-activating agents NGF (100 ng/ml) or basic fibroblast growth factor (100 ng/ml). As seen in Fig. 5G, exposure to PACAP, forskolin, or 8-CPT-cAMP caused growth arrest in NS-1 cells or Epac1-deficient NS-1 cells, whereas Epac2-deficient cells proliferated at a similar rate to untreated cells under these conditions. NGF and basic FGF caused growth arrest in all three cell lines tested. Together, these data indicate that although Epac2 is necessary for cAMP-dependent growth arrest in NS-1 cells, Epac1 is dispensable for this process.  , lanes 4 -6). The abundance of Epac1 and Epac2 protein measured in untransduced NS-1 cells (lanes 1 and 4) was compared with NS-1 cells stably expressing shRNA against Epac1 (shEPAC1, lanes 2 and 5) or Epac2 (shEPAC2, lanes 3 and 6) To quantitatively monitor cell cycle arrest, we employed PC12 cells transduced with LNCX-HDHBc-tdimer2, a retroviral vector encoding a G 1 /G 0 biosensor fused to red fluorescent protein that distinguishes proliferating cells from growth-arrested cells by confining red fluorescent protein expression to the nucleus in growth-arrested cells, whereas it is diffusely expressed in the cytoplasm in cells in S/G 2 phase (23,37). Treatment with either 8-CPT-cAMP (100 M) or 8-CPT-2Ј-O-Me-cAMP (100 M) for 48 h caused growth arrest in the majority of PC12 cells (Fig. 6B). Furthermore, as seen in blocked PACAP-dependent growth arrest while not affecting neuritogenesis. Growth arrest without neuritogenesis was observed in cells treated with 8-CPT-cAMP (100 M) and SQ22,536 (1 mM), most likely because of the ability of cAMP to activate Epac, but not NCS/Rapgef2, under these conditions. Together, these data indicate that Epac is responsible for cAMP-induced growth arrest in PC12 cells while not being involved in neuritogenesis.
NGF-and cAMP-dependent Signaling Pathways both Require p38 to Induce Growth Arrest in NS-1 Cells-It has long been appreciated that trophic factors such as NGF cause growth arrest in PC12 cells (5). However, it has not been determined whether cAMP and NGF cause growth arrest by parallel activation of a common target or by cross-activation of their respective signaling pathways. In NS-1 cells, treatment with NGF for 3 days also induced growth arrest that was sensitive to SB 239063 (Fig. 7A). We next measured NGF-induced p38 activation, and, as seen in Fig. 7, B and C, both NGF and 8-CPT-2Ј-O-Me-cAMP caused significant increases in p38 phosphorylation. Because Ras activation is one of the best characterized signaling effectors of NGF for differentiation (38), we cotreated NS-1 cells with NGF and FTS (10 M), a specific Ras inhibitor (39,40). FTS significantly inhibited NGF-dependent p38 activation, whereas Epac-induced p38 activation stimulated by treatment with 8-CPT-2Ј-O-Me-cAMP was insensitive to FTS (Fig. 7, B and C). Both NGF (100 ng/ml) and 8-CPT-2Ј-O-Me-cAMP (100 M) caused growth arrest (Fig. 7D). The effects of NGF were selectively blocked by the addition of 10 M FTS or 200 nM K252-a (Fig. 7D), neither of which affected growth arrest induced by 8-CPT-2Ј-O-Me-cAMP. In the same experiments, ESI-09 (10 M) significantly inhibited growth arrest elicited by 8-CPT-2Ј-O-Me-cAMP while not affecting NGF-dependent growth arrest (Fig. 7D). These data indicate that p38 activation is necessary for growth arrest caused by either NGF or cAMP. However, NGF activates p38 through a Ras-dependent pathway, whereas cAMP signals through an Epacdependent, Ras-independent pathway.
Protection of NS-1 Cells from Serum Withdrawal-induced Cell Death by cAMP Requires PKA Activation-Along with causing neuritogenesis and growth arrest, cAMP has been shown to be cytoprotective in multiple cell types, including PC12 cells (41). We wished to determine whether cAMP-dependent cytoprotection is similarly parcellated to a single cAMP sensor, as growth arrest and neuritogenesis are, or whether cytoprotection mediated by cAMP requires signaling through multiple cAMP sensors. NS-1 cells were cultured for 3 days in serum-free medium to induce apoptosis (42), and then cells were stained with PI, a highly polar compound that fluoresces upon binding to DNA and, thus, only stains cells with damaged plasma membranes, i.e. dead or dying cells. Serum withdrawal caused the death of most of the cells (Fig. 8), whereas only 1-4% of control cells grown in serum incorporated PI. As seen in Fig. 8, A and B, treatment with 8-CPT-cAMP (100 M) significantly decreased the ratio of PI-positive cells after serum withdrawal, thus protecting NS-1 cells from apoptosis. Cyclic AMP-dependent survival in serumfree medium was blocked by H-89 (30 M) while remaining unaffected by ESI-09 (10 M) or SQ22,536 (1 mM), suggesting that survival may be PKA-dependent. Consistent with this notion, we found that PACAP-38 (100 nM) also supported cell survival in serum-free medium (Fig. 8, C and D), and PACAP-dependent survival was blocked by the addition of a specific antagonist of cAMP binding to the regulatory subunit of PKA, Rp-8-Br-cAMPS (43), applied at 750 M (Fig. 8,  C and D). Furthermore, a specific agonist of the regulatory subunit of PKA, Sp-8-Br-cAMPS (500 M), protected NS-1 cells from serum withdrawal (Fig. 8, C and D). Together, these data suggest that cyclic AMP-dependent survival is mediated exclusively by the activation of PKA and does not require Epac or NCS/Rapgef2.
As expected, NGF also protected NS-1 cells from serum withdrawal-induced cell death (Fig. 8, E and F). However, NGFmediated rescue from serum withdrawal-induced cell death was not blocked by Rp-8-Br-cAMPS, indicating that NGF does not protect cells by activating PKA and suggesting that cAMPand NGF-dependent signaling may share a final common target for prosurvival signaling but do not share upstream signaling components for such signaling.
Taken together, these data (summarized in Fig. 9) suggest that cAMP and NGF trigger distinct but coordinated differen- tial processes by activation of multiple parallel signaling pathways. Furthermore, we show that cAMP-dependent neurite extension, growth arrest, and cytoprotection are mediated by distinct pathways, each gated by a separate cAMP sensor: NCS/ Rapgef2, Epac, and PKA, respectively.

DISCUSSION
In NS-1 cells, neuropeptide-induced cyclic AMP elevation and neurotrophin-induced TrkA activation are both sufficient to cause differentiation, as defined by the parallel cellular processes of neuritogenesis and cell growth arrest. Our data are consistent with the widely held model that NGF signaling for differentiation is Ras-dependent and show that neuritogenesis requires ERK but not p38, whereas growth arrest requires p38 but not ERK activation. When initiated by NGF, neither of these processes require cAMP. Neuropeptide-induced differentiation, likewise, requires activation of ERK for neuritogenesis and p38 for growth arrest, but (at least for PACAP) these pathways were differentially activated by the cAMP sensors NCS/Rapgef2 and Rapgef4 (Epac2), respectively.
The Epacs, identified as Rap1 GEFs (6,7), are members of the Rap GEF gene family (44) and enhance Rap1 activation in NS-1 cells. However, our results suggest that Rap1 activation is not a requisite step for Epac-dependent p38 activation. In fact, other Rap-independent functions of Epac have been noted. For example, Epac has been shown to activate the MAP kinase JNK via a Rap-independent mechanism in HEK 293T cells (45). Epac has also been shown to be an indirect activator of the small G protein Rit (11), which, unlike Ras or Rap, does not require prenylation for its biological activity (46). Interestingly, Epac1-Rit-p38 signaling has been reported to be a component in PACAP-induced differentiation of PC6 pheochromocytoma cells (11). Rit or a related small G protein may be involved in the Epac2-dependent p38 activation reported here in NS-1 cells, especially in light of the fact that an inhibitor such as FTS failed to interfere with Epac-dependent p38 activation while completely blocking NGF-dependent p38 phosphorylation.
Our finding that Epac2, but not Epac1, mediates cAMP-dependent growth arrest may shed important light on the distinct roles of these two Rap GEFs on cell cycle regulation in systems other than neuroendocrine cells. For example, it is not known whether p38 mediates effects of either Epac1 (Rapgef3) or Epac2 (Rapgef4) on cell proliferation or growth arrest in nonneuroendocrine cells. In fact, the Epacs have been reported to both positively and negatively affect cell cycle entry and exit, depending on the cell type and relative expression levels of Epac1 and Epac2 (47). P38 activation has been shown to exert antiproliferative effects in many cell and tissue types, including tumors in skin, lung, and liver (48), and p38 hypoactivity has been noted in human tumors (32). In this study, Epac-dependent activation of the p38 pathway was shown to be necessary for growth arrest of NS-1 pheochromocytoma cells, and, therefore, it is possible that Epac/p38 signaling may play an inhibitory role in the development of neuroendocrine tumors. In general, at least on the basis of the current limited examples, it would appear that Epac exerts growth arrest through one of several MAP kinases in cells in which differentiation and growth arrest occur in parallel and that Epac exerts a proliferative effect, through other MAP kinases, in cells in which differentiation is linked to proliferation or at least not to obligate removal from the cell cycle.
The neuritogenic and growth arrest pathways activated by cAMP in NS-1 cells require the cAMP sensors NCS/Rapgef2 and Epac, respectively, and exclusively (see above). Remarkably, there is no apparent role for the classical cAMP sensor/effector PKA in either growth arrest or neuritogenesis in neuroendocrine cell differentiation. However, the prosurvival effect of cyclic AMP after serum withdrawal in NS-1 cells was blocked by a PKA inhibitor but not by Epac or NCS/Rapgef2 inhibitors. Furthermore, the prosurvival effects of PACAP are sensitive to a specific antagonist of the regulatory subunit of PKA and are fully mimicked with the PKA-specific cAMP analog Sp-cAMPS. These data imply that PKA is the sole sensor mediating these effects of PACAP with no apparent contribution from the other two cAMP sensors, NCS/Rapgef2 and Epac.
Ligands of G s -coupled GPCRs, notably PACAP, have also been shown to exert PKA-dependent cytoprotective effects in numerous cell types, including neurons of the central nervous system and retina, cardiomyocytes, kidney cells, and hepatic cells (49 -52). NGF also protected cells from death following serum withdrawal. However, in contrast, NGF-mediated cytoprotection was PKA-independent, which is consistent with reports that NGF-mediated survival effects require the activation of phosphatidylinositol 3-kinase and Akt (53,54).
In addition to cAMP signaling specificity conferred by differential expression of cAMP sensors in different tissues, it is evident that cAMP sensors are activated to perform distinct and specific tasks within the same cell following cAMP elevation after stimulation of G s -coupled GPCRs. Furthermore, although stimulation of any G s -coupled GPCR is presumed to lead to the activation of PKA, a subset of G s -coupled GPCRs, including PAC 1 (8) and ␤ 1 -adrenoceptors (55), efficiently activate NCS/ Rapgef2, whereas others, such as ␤ 2 -adrenoceptors, do not (8). It is an intriguing possibility that these differential signaling properties could be leveraged to specifically regulate a single aspect of cAMP-dependent signaling, such as morphological plasticity, proliferation, or cell survival, in vivo.
A further question of interest is whether biased GPCR ligands that lead to preferential activation of particular cAMP sensors can be identified. In fact, different structural analogs of NGF have been shown to differentially induce signaling, leading to either differentiation or trophism/cytoprotection through TrkA receptors (56,57). GPCR ligands that specifically modulate these cellular outcomes in a biased fashion (58) could, therefore, be of great therapeutic potential.